Announcing AI Services v0.2.0

We are excited to announce the release of AI Services v0.2.0, part of the Open-Source AI Foundation for Power, designed to accelerate AI adoption on IBM Power11 systems with the IBM Spyre® Accelerator.

AI Services provides a streamlined way to deploy and run AI workloads on Power, including Retrieval-Augmented Generation (RAG) applications and document processing pipelines. With each release, we continue to simplify deployment, expand capabilities, and improve operational visibility for enterprise environments.

What’s New in v0.2.0

This release introduces enhancements across deployment flexibility, document processing, and observability.

AI Services now includes OpenShift runtime support, enabling deployments on Red Hat OpenShift with automated bootstrap for Red Hat OpenShift AI (RHOAI) and Spyre operators. Helm-based deployment templates and runtime-aware lifecycle management make it easier to manage applications consistently across environments.

Language capabilities have been expanded with support for processing and querying German documents, enabling more flexible handling of multilingual enterprise data.

A new Digitize UI has been introduced to simplify document workflows. This interface allows users to ingest documents used by the Q&A service and digitize PDF documents, making them searchable and ready for downstream AI applications.

The release also introduces a Summarization API, enabling fast, high-quality summaries for text, TXT, and PDF files. This capability supports use cases such as document review, summarization, and knowledge extraction.

To improve automation and integration, an API server for the ingest-CLI service is now available. It provides endpoints for asynchronous ingestion and document digitization, enabling seamless integration with external systems and workflows.

For environments without accelerators, a CPU-only RAG template has been introduced. This allows users to deploy and test AI workloads in resource-constrained or cost-sensitive environments while maintaining architectural consistency.

On the storage side, AI Services has migrated from Milvus to OpenSearch as the vector backend, enabling improved scalability, unified search capabilities, and better integration with enterprise ecosystems.

Finally, this release enhances observability by introducing performance metric collection in the Q&A API server along with log traceability across all services, providing better visibility into system behavior and simplifying troubleshooting.

Getting Started

To explore AI Services and get started:

• Documentation: https://ibm.biz/aiservices

• GitHub: https://github.com/IBM/project-ai-services